Active performance monitoring metrics aren’t just for finding problems

Using active monitoring to manage network capacity more effectively

Network performance monitoring is becoming increasingly important for modern networks. This is particularly true of 4G and will be crucial to driving revenue and reducing risk as we march towards 5G. Having accurate and granular metrics that address quality of service (QoS) by showing latency, packet delay variation (PDV) and packet loss are critical for identifying many issues such as voice over LTE (VoLTE) performance and congestion and path changing problems.  

However, there is another powerful use for these metrics which takes advantage of active monitoring’s granularity and percentile values. By providing greater insight into what is actually happening on the transport link — beyond minimum, average, and maximum values — operators can greatly improve network performance.

Running the network “red hot”

Given the high cost of infrastructure and declining ARPU, operators cannot afford to have under-utilized assets in their network. As the CTO of a Tier 1 service provider told me recently, they want to run the network “red hot”.

So, instead of just using performance monitoring metrics to identify issues and problem areas, it is also possible to apply them in spotting areas that have spare capacity. 

Traditionally, capacity might be looked at using such things as SNMP to poll routers for network utilization every 15 minutes or so. However, this only gives you an average value and may hide the true peak utilization for the circuit. For example, two similar links may both show 30% utilization during a busy hour. In reality, one of these links may indeed have a relatively flat traffic profile – that is typically around the 30% mark. But the other link may be constantly bursting to close to 100% for short periods of time and then dropping back to a very low sustained traffic rate. Both links will show up as 30% average utilization, but the customer experience would really be quite different.

Going granular with packet delay variation

One of the ways that we can differentiate between these links is by looking at packet delay variation (PDV). PDV measures the delta between the lowest latency packet for the sample period and the highest. For the above example, we can say the following:

1. For the link that has a relatively consistent 30% utilization, the PDV value will be exceedingly small. Basically, no packets are being impeded due to congestion and the various latency percentile values will be very similar to the average value. 
2. For the second link suffering from high burst values, buffers could be filling up on aggregation routers, causing some packets to be delayed. This will result in a higher PDV value – perhaps 10 to 20ms or more. The percentiles are important here, as we are not as concerned about the maximum value of PDV (just one packet out of a large sample size) as we are about how many, or what percentage of packets, are being impacted.

Managing capacity with smarter performance management

Now we can use the key performance indicator (KPI) of PDV to identify which sites during peak times are not suffering from congestion-related performance degradation. If the PDV is relatively low (less than 1ms) across the course of a typical day, then it is fairly safe to say that there is available capacity on that link.

With this knowledge, we can identify which sites in high-value areas like CBDs are congested and which have spare capacity. Thus, design decisions can be made around routing and ring configurations, which macro sites can be used to act as hubs for small cells, and even which links would benefit from a capacity upgrade (from 1G to 10G, for example). If some load can be shifted from the congested sites to the sites with available capacity, perhaps via radio cell configurations, we can get closer to the goal of running the network “red hot”.

That’s good news for both network operators and their customers.

To see how active monitoring can help you manage network capacity more effectively, book a demo with Accedian’s performance experts here.